For much of manufacturing’s modern history, the quality of Manufacturing Execution Systems (MES) has set the boundaries for achieving innovation. An MES captures data from equipment and, to some extent, from people to measure the execution process. What is measured gets done, but the trouble is that MES’s have historically been mediocre or at least inflexible. They have ballooned in features, requiring specialized expertise. Over time, they have limited the scope for innovation and change rather than extending it. Are there next generation MES’s on the horizon that promise to improve the situation drastically? Yes. And slowly but surely, they are driving the entire set of vendors towards simplicity.
The Emergence Of MES
Industrial tech has been an endless evolution of systems that do not always work well together. But as industrial processes have evolved so too have the systems. Manufacturing execution systems (MES) is a recent label taking into account the important operational functions of these systems.
According to Gartner’s Magic Quadrant methodology, MESs are computerized systems used to track and document the transformation of raw materials to finished goods. The 2022 Gartner Magic Quadrant for MES, which they made available to clients on June 1, is available for download on various vendor web sites this year with dedicated press releases that indicate they made the list (alphabetically): Apprentice.io (PR), Critical Manufacturing (PR), GE (PR), iBASEt (PR), Rockwell (PR), Tulip (PR), although many that made similar previous announcements in previous years don’t yet have a press release, e.g. Honeywell (2021 PR) and iTAC (2020 PR), for example.
Most “Best of” (Gartner, Selecthub) or long lists of MES that at times include ERP systems and others (Capterra, G2, Slashdot, Softwareconnect, Sourceforge, Thomasnet, Trustradius) include legacy vendors such as Aveva (1967), Parsec (1987), Dassault Systèmes (1981), GE Digital (2015), Oracle (1977), Rockwell Automation (1903), Siemens Digital (2007), Honeywell (1906), Critical Manufacturing (2009), Körber Werum (2014) and AVEVA (2017) and a few key lists include newcomers such as Pico (2019) and Tulip (2012).
In short, all of these systems manage, monitor, and synchronize the execution of real time, physical processes involved in manufacturing operations. The legacy systems are responsible for tracking and tracing the operational activities of vast arrays of factories and processes. The newcomers do so with increased attention to user interfaces while benefiting from the latest technology without the legacy tie-ins. Though obviously without the benefit of decades of industrial experience as their operating vendors.
An MES captures the manufacturing process’s data, processes, and outcomes, both for regulated and unregulated industries, with numerous benefits, including improving efficiencies, sharing best practices, reducing downtime, and increasing the consistency and safety of manual processes. Often, this is achieved through enabling paperless and, to a large degree, uninterrupted workflow, which maintains a worker’s concentration and efficiency during crucial procedures. Another key outcome is the near eradication of just-in-case inventory.
I don’t know if a readable introduction to MES systems exists. Seldom has there been such a misunderstood, misconstrued, and mislabeled piece of software as fundamental as MES. What do I mean? The essence of such systems should be (and often is) wider than manufacturing. The system in question always aims for more than “execution” (whatever that means). Lastly, they are rarely systems but bundles consisting of usable and non-usable parts. Each part is typically too big and expensive to handle. As a result, a vast market for systems integrators and consultants has evolved (ControlEngineering, MarketsandMarkets, Mordor Intelligence, Verified Market Research).
That said, an MES should excel at shop floor management across a wide range of industrial use cases, and should, optimize the use of shop floor resources, assist with scheduling minute to minute (not just by day or shift), and should at the very least be an excellent “system of engagement” where it ties up loose ends for mobile workforce along assembly lines, shop floors, or across factories. A MES is typically optimized for complex production processes with frequent bottlenecks (the bane of production lines), limited capacity (a common occurrence), and as a way to deal with alternative paths in production. It should also be a “system of record” at times, although that role is typically better played by an ERP system. One way of looking at it is by considering that MES software is generally used for the “Who, What, When, Where, Why and How Much” facets of work order execution (Lamb 2015).
Where Are The Humans?
MES, by its very design, is not human-centric. Worse, it disconnects work from the floor where it occurs. MES solutions typically have a logic built into predefined modules.That means that the system predetermines certain ways of doing things, and the shop floor has to adapt. Flexibility suffers when abrupt changes occur because the MES lacks frontline workflows that require programming experience to configure. Even programmers would butt their head against the rigid pre-built modules and data structures that were intended to give stability. They might collect data from (big, expensive) machines but will not collect (so much) data from humans or perhaps not from more peripheral custom devices and sensors because they were not built into the first installation.
Furthermore, data analysis might happen “next day, next week” instead of in real time–the way newer machine monitoring point solutions do. Lastly, the biggest issue for many customers is evidently the expensive upfront installation, site license, the system integration services that follow, and the annual contract based on factory floor headcount. Half the fight is to change management’s perception of what they are trying to accomplish. Control is never the desired outcome. Deeply understanding what has to be done, comparing that to what is being done, and widening the scope of work so workers can thrive and excel, is.
It is not an industry secret that manufacturers are frustrated with traditional MES systems. A plethora of point solutions have emerged to try to fix parts of the problem. There are one trick ponies for machine monitoring (e.g., companies such as MachineMetrics). GE tried to build a much more ambitious solution with GE Digital’s Predix but failed despite massive investments perhaps because software is seldom built by cash infusion absent a clear path of user-centric development (Kumar 2019; Mann and Gryta 2020). Either way, you find connected worker solutions that can help digitize workflows but don’t necessarily tie in to all other systems (Dozuki, eFlex Systems, L2L, Parsable, Poka, VKS). There are generic low-code or no-code platforms that include manufacturing among their target markets but with generic digital solutions (Appian, Mendix). Additionally, industry specific solutions have emerged, for example for life science (Emerson, Kneat, Körber Werum, MasterControl, Valgenesis), food, beverage and CPG manufacturers (Redzone), asset intensive industries (Arundo, Cognite), or automotive (Workclout). Lastly, business function specific solutions also now exist, such as auditing (iAuditor).
Must Manufacturing Execution Systems Turn Mastodonic?
Discrete manufacturing involves assembling and making distinct things such as manufacturing cars, bicycles, or mobile phones. Process manufacturing involves mixing ingredients according to specific formulas or recipes, such as paints, specialty chemicals, textiles, or pharmaceuticals. Traditional software for manufacturing has always been dedicated to one of these industry modalities. Rarely have software vendors been able to support both discrete and process manufacturing without complicated customizations and workarounds. However, even so, these software systems seem to quickly grow in complexity, license and site costs, and require specialists to implement and service.
If you look at the top ten industrial automation companies in the world, which includes Siemens, ABB, Emerson, Rockwell Automation, Schneider Electric, Mitsubishi Electric, Yokogawa Electric, Omron Automation, and Danaher Industrial Ltd. (Plant Automation), many of these companies have either acquired an MES system or have contributed to the success of the space by providing enormous contracts to these systems. This makes sense, from their point of view, if you think about the complexity of the products they make and the effort it takes to manufacture them. However, their approach slows industrial innovation because it keeps dinosaurs alive, and the marketplace suffers inferior systems as a result.
A MES is, typically, a rigid system and takes time to customize. According to Gartner, the average implementation time for an MES is 15-16 months (Tulip 2021); others estimate 6-18 months (Swanton and Smith 2005). Custom-built MES configurations based on MES tool kits can run high ratios of license-to-service dollars, often upwards of 1:5. That means for every $10,000 spent on licenses, you may actually be spending $50,000 in services. The total cost of ownership of custom-built or off-the-shelf MES is high (Tulip 2021). There is near universal frustration with MES systems but up until recently there have been few realistic alternatives.
However, a MES does not need to be mastodonic. The ISA-88 and ISA-95 standards do make data transfer between systems feasible. Features can improve faster. Any system that needs to be on premise, misses the opportunity to natively and flexibly rely on cloud computing. The big systems that resulted from solutions being patched together over time could be ending. In fact, I predict startups currently providing point-solutions to eventually merge with more complete offerings that have similar functionality without the same complexity. It does not hold the test of agility, especially in build-to-order, high-mix-high-volume manufacturing.
MES in its current form is dying. Partly, the infrastructure it is based on is getting old. The Purdue Enterprise Reference Architecture (PERA), a structural model for industrial control system security (e.g., “The Purdue model”) was developed back in 1990, before the internet and cloud computing took off and will need to be complemented with the flexibility of contemporary Edge systems (Greenfield 2020). The key challenges of Industrial IoT include standardization, connectivity across vendors, and defining data models and digital twins across big industry verticals including automotive and manufacturing. That work is underway.
Legacy MES vendors are moving to the cloud. In principle, this is goodness. In reality, some of the older solutions don’t stand up so well against cloud requirements and are scarcely able to reap the benefits of a fully web-native solution because they were coded in a different age. All that legacy code either sticks with the solution or it has to be abandoned at great cost. A typical strategy deployed by legacy vendors is to acquire newer solutions and patch them on top of their old one. That can work, but does result in a patchwork of code and rarely furthers the simplicity agenda.
Emerging MES – New Directions
If the concept of MES is getting old, what are newer technologies and firms doing about it? I’ll give three examples: Pico is developing MES for small-and medium enterprises. Biophorum has developed a MES of the Future Manifesto to answer the challenge from the emerging field of biomanufacturing. Tulip has built a frontline operations platform that bypasses several of the obstacles presented by current MES’, featuring less reliance on software experts, and providing a cloud-native platform that installs, runs, and scales easily.
Although 98.6% of manufacturers in the U.S. have fewer than 500 employees, these companies produce over $3 trillion of goods each year (see Pico MES is creating smarter, smaller factories). Pico’s software is “priced, configured and optimized to meet the needs of these smaller manufacturers” and smaller factories. Pico’s pre-built library of tools and machine integrations and low-cost IoT devices make installation easier and even includes a digital twin to compare and contrast ideal and actual machine performance. Pico CEO Ryan Kuhlenbeck says to VentureBeat: “Legacy systems were designed for the complexities of the large manufacturers, making them expensive, difficult to use and requiring specialized support personnel,” (VentureBeat 2022).
Biophorum, a trade association type network for the emerging biomanufacturing industry, which encompasses activities of several known life science companies, uses the Swiss Army knife analogy about MESs; designed to be near universally applicable and with a lot of unnecessary functionality. In contrast, they wish for a “flexible, scalable and process agnostic” MES that is “easily maintainable” and contains “libraries of biomanufacturing processes and operations.” (see Welcome to the MES of the Future). Biophorum points to the new manufacturing modalities that combine various types of batch manufacturing at different scales from single patient to clinical and with variable needs for containment. Notably, this applies to precision medicine where a batch of one means the architecture of production is radically altered. This kind of extreme customization is hardly possible with a current generation MES.
Tulip’s frontline operations platform was designed with the industrial engineer in mind, aiming to reduce the need for software competence and skills when building industrial applications that track production-related activity on the shop floor or beyond. The subscription-based business model allows for quick ramp-up, experimentation, and tailoring to each site and use case. However, the real differentiator is that a frontline operations platform can solve challenges that traditional MES cannot. By rapidly connecting the people, machines, devices, and the systems on a shop floor or along a supply chain, creating a seamless value chain built by the organization itself, and its citizen developers, not by vendor decree, the solution creates workforce engagement rather than garnering the typical complaints leadership gets from workers when implementing a new IT solution.
The Future of MES is Augmented Lean
Forrester recently predicted that manufacturers will double down on creativity and sustainability in the years to come (see Predictions 2022: Top-Tier Manufacturers Will Prioritize Creativity And Sustainability). Machines and systems are not typically creative, but a human workforce can be. Whatever manufacturers choose to do, if they want to change, they need to be more flexible.
The future of MES is a leaner application that does not rely on specialists to implement. That way, it can be further customized with ease at any point in time, which matches not only biomanufacturing, but small-and medium enterprises, and ostensibly, most emerging manufacturers and production lines. As most businesses are shifting towards more human-centric operations, it helps to have technologies that serve that purpose. However, it is not sufficient. Managing industrial change requires a mindset that combines hacking and governance, which my co-author Natan Linder and I call Augmented Lean. In good Lean tradition, it is better to augment a process or a worker than try to automate it away, as long as the end result is simplicity, because complexity leads to waste, and having a workforce (in addition to a machine park) has a value in itself. Ideally, systems, sensors, workers each and all, separately, and together, contribute to operational simplicity. Anything else undoubtedly takes us in the wrong direction.
This is an abridged excerpt (with permission from the publisher) from a forthcoming book by Natan Linder and Trond Arne Undheim, entitled Augmented Lean: A Human-Centric Framework for Managing Frontline Operations, Wiley, 2022.
Disclaimer: I have financial interests in Tulip, referred to in this article.
How Cotopaxi Provides Sibly’s AI-Assisted Coaching 24/7 To Employees
Irish Drone Delivery Startup Expanding To U.S. And Mainland Europe
What Brands Need To Know: Marketing Measurement And Attribution In 2023